Page 71 - AI for Good-Innovate for Impact - An Interim Report 2024
P. 71

AI for Good-Innovate for Impact



               Use Case – 14: Innovative Practices of Cloud Networking Enabled

               by Network Large Model                                                                               14-ChinaTelecom













               Country: China

               Organization: CHINA TELECOM

               Contact person: LYU Tiantian lvtt@ chinatelecom .cn


               14�1� Use case summary table

                Domain            5G/6G
                Problem to be     The current manual assistance is unable to meet the challenges of opera-
                addressed         tion and maintenance.
                                  It is necessary to start from the requirements of the operator’s network,
                                  including user service requirements such as Accurate business recom-
                                  mendations, Network operation and maintenance requirements such
                                  as Network fault frequency, slow response,  use large models to build
                                  an architecture that adapts to the complex cloud-network operations of
                                  the future、to meet the high-order evolution of self-intelligence require-
                                  ments..

                Key aspects of the  The Qi Ming network large model will employ the advanced structure
                solution          and algorithm to handle the complexities. It will also integrate network
                                  knowledge, business, and operational data through large-scale pre-train-
                                  ing, fine-tuning, and optimization.

                Technology        Network large model and
                keywords          AI agent technology


                Data availability  Private data: we use some sensitive words from
                                  https:// github .com/ jkiss/ sensitive -words, but do some processing in the
                                  combination with telecom data features.

                Metadata (type of   China Telecom’s massive network data and knowledge corpora.
                data)             Text data, Logs, alarms, and configurations

                Network           Professional intent identification, ubiquitous generation, logical orches-
                Large Model       tration, collaborative scheduling, predictive analytics
                Characteristics

                Testbeds or pilot   Inside China Telecom
                deployments











                                                                                                     57
   66   67   68   69   70   71   72   73   74   75   76